REVIEW

Artificial Intelligence, Clinical Decision

Support Algorithms, Mathematical Models,
Calculators Applications in Infertility:
Systematic Review and Hands-On Digital
Applications
Carlo Bulletti, MD; Jason M. Franasiak, MD; Andrea Busnelli, MD;
Romualdo Sciorio, BSc, Msc; Marco Berrettini, PhD; Lusine Aghajanova, MD, PhD;
Francesco M. Bulletti, MD; and Baris Ata, MD

Abstract

The aim of this systematic review was to identify clinical decision support algorithms (CDSAs) proposed
for assisted reproductive technologies (ARTs) and to evaluate their effectiveness in improving ART cycles
at every stage vs traditional methods, thereby providing an evidence-based guidance for their use in ART
practice. A literature search on PubMed and Embase of articles published between 1 January 2013 and 31
January 2024 was performed to identify relevant articles. Prospective and retrospective studies in English
on the use of CDSA for ART were included. Out of 1746 articles screened, 116 met the inclusion criteria.
The selected articles were categorized into 3 areas: prognosis and patient counseling, clinical management,
and embryo assessment. After screening, 11 CDSAs were identified as potentially valuable for clinical
management and laboratory practices. Our findings highlight the potential of automated decision aids to
improve in vitro fertilization outcomes. However, the main limitation of this review was the lack of
standardization in validation methods across studies. Further validation and clinical trials are needed to
establish the effectiveness of these tools in the clinical setting.
ª 2024 THE AUTHORS. Published by Elsevier Inc on behalf of Mayo Foundation for Medical Education and Research. This is an open access article under
the CC BY license (http://creativecommons.org/licenses/by/4.0/) n Mayo Clin Proc Digital Health 2024;2(4):518-532

T
he global demand for assisted repro- support algorithms (CDSAs) and artificial in-
ductive technologies (ARTs) has telligence (AI), including machine learning
From the Help Me Doctor, increased substantially, although de- (ML), has become prominent. These tools are
Assisted Reproductive Tech-
nology, Gynecological Endo-
livery rates from these procedures remain designed to aid clinicians and patients in mak-
crinology and Reproductive moderate, with about 25% success per autolo- ing informed decisions by managing and inter-
Surgery, Cattolica, Italy (C.B.); gous embryo transfer and 30% for frozen- preting vast, complex data sets.4,5 When
Department of Obstetrics,
Gynecology, and Reproduc-
thawed transfers, according to the European validated, they hold potential to enhance clin-
tive Science, Yale University, IVF Monitoring Consortium for the European ical prognosis, diagnosis, and treatment.
New Haven, CT (C.B.); IVI Society of Human Reproduction and Embry- Furthermore, they can empower patients in
RMA, Reproductive Medicine
Associates of New Jersey, NJ
ology.1 Despite this, cumulative live birth rates making informed decisions by reducing uncer-
(J.M.F.); Department of improve with multiple cycles, but repeated tainty, thereby fostering a new era of personal-
Biomedical Sciences, Human- treatment and extended duration can exert ized medical care.4,6,7 However, the misuse or
itas University, Milan, Italy
(A.B.); IRCCS Humanitas
emotional and financial strains.2,3 misinterpretation of these systems can lead to
Research Hospital, Rozzano, Assisted reproductive technology is a com- inappropriate interventions or misinformed
Italy (A.B.); Fertility Medicine plex process, involving various techniques patient expectations. Overall, despite the pro-
and Gynaecological Endocri-
nology Unit, Department
such as ovarian stimulation (OS), in vitro liferation of such tools, their actual benefit in
fertilization (IVF), and embryo cryopreserva- improving ART outcomes remains to be fully
Affiliations continued at
the end of this article. tion. Recently, the use of clinical decision established.6,8e10 This review aimed to

518 Mayo Clin Proc Digital Health n December 2024;2(4):518-532 n https://doi.org/10.1016/j.mcpdig.2024.08.007

www.mcpdigitalhealth.org n ª 2024 THE AUTHORS. Published by Elsevier Inc on behalf of Mayo Foundation for Medical Education and Research. This is an open
access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
CLINICAL DECISION SUPPORT SYSTEMS IN INFERTILITY

critically evaluate these tools and categorize

them based on evidence. ARTICLE HIGHLIGHTS

d In recent years, several clinical decision support algorithms for

METHODS assisted reproductive technologies have been proposed with the
A systematic review was performed to identify promise to improve treatment outcomes.
CDSA, AI applications, algorithms, mathemat-
d Automated decision aids may be promising to enhance IVF
ical models, and calculators that have been
proposed to support physicians and embryol- outcomes, whereas the lack of standardization in validation
ogists at different steps of ART, namely predic- methods across studies is a key limitation.
tion of outcomes, patients profiling, treatment
optimization and personalization, embryo
evaluation, and laboratory management. The searched bibliography of included articles
protocol for this systematic review has been along with screening similar articles.
registered with the international prospective
register of systematic reviews PROSPERO
Selection of Studies and Data Extraction
(CRD42024499571). The systematic review
After automated deletion of duplicates, 2 re-
was conducted according to the PRISMA
searchers (C.B. and R.S.) independently
guidelines.11
screened titles and/or abstracts of the studies
retrieved. Additional sources were screened
Search Strategy to identify studies that potentially met the in-
We searched PubMed and Embase for articles clusion criteria. The full text of these poten-
published from 1 January 2013 to 31 January tially eligible studies was retrieved and
2024. Prospective and retrospective studies in independently assessed for eligibility by 2
English on the use of CDSA for ART were members of the review team. Any disagree-
included. Specific keywords included the ment between them over the eligibility of
following: assisted reproduction, medically assis- any study was resolved through discussion
ted reproduction, in vitro fertilization, IVF, em- with a third reviewer (M.B.). Prospective and
bryo culture, AI, ML, prediction models, retrospective studies and randomized
automated algorithm, calculator, deep learning, controlled trials were considered eligible for
automatic detection, automated data processing, inclusion. Case reports and abstract only re-
time-lapse, and ultrasound. Inclusion and ports were excluded. Two collaborators (D.C.
exclusion criteria are documented with the pa- and D.F.) extracted the data using the stan-
tient/population, intervention, comparison dardized tables developed by the authors to
and outcomes framework (Table 1). To sup- ensure consistency. Studies comparing auto-
plement the database searches, we hand- mated models vs humans were further

TABLE 1. Patient/Population, Intervention, Comparison and Outcomes Model of the Systematic Review
Conducted.
Population Individuals or couples undergoing medically assisted reproduction procedures
Intervention Computer-based decision aid systems designed to improve assisted reproductive technology
outcomes at any stage
Comparison Nonautomated systems (eg, human observation/nonautomated decision-making systems)
Outcomes i. Implantation rate
ii. Clinical pregnancy rate
iii. Live birth rate

Study type Inclusion criteria: interventional studies, prospective and retrospective studies, observational
studies, and descriptive studies
Exclusion criteria: case reports, editorials, letters to the editors, abstracts only
Date From 1 January 2013 to 31 January 2024
Language English

n n
Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007 519
www.mcpdigitalhealth.org
 MAYO CLINIC PROCEEDINGS: DIGITAL HEALTH

Records identified through Records identified through

Identification
database searching: PubMed hand searching
(n=1769) (n=10)

Records removed automatically

(n=33)
Screening

Titles/ abstracts screened Reports excluded

(n=1746) (n=1537)

Full-text articles assessed Reports excluded:

for eligibility Out of scope (n=80)
(n=209) Weak evidence (n=13)
Included

Studies included in qualitative synthesis

(n=116)

FIGURE 1. Study flow chart according to PRISMA guidelines.

evaluated for statistical significance. Results 1. A noninvasive early presumption diag-

were considered significant when P<.05. nosis method for pelvic endometriosis
with a diagnostic penetrance of 90%.12
2. A calculator to estimate the ovarian
RESULTS oocyte reserve based on the anti-Mülle-
The literature research identified 1746 articles rian hormone (AMH), and/or antral folli-
that were further screened with the 2 progres- cle count, and based on previous
sive selection steps to identify articles that met response to OS able to identify patients
the inclusion criteria (Figure 1). A total of 116 with potential poor or suboptimal
studies were included in the final analysis response to OS according to the Bologna
(Figure 1). Articles identified were divided ac- or POSEIDON criteria.13,124
cording to the different stages of ART 3. Artificial intelligence for sperm count with
(Table 2).6,8e10,12e123 evaluation of motility and morphology to
classify patients as normal, hypospermic,
and azoospermic based on published
Evidence in favor of CDSA in ART criteria60,61,125 and to assess DNA
We identified 11 automated tools that have re- fragmentation.126
ported efficacy for patient prognosis and coun-
4. A noninvasive tool to diagnose polycystic
seling and clinical management as follows:.
ovary syndrome (PCOS) based on body
d CDSA for patient prognosis and counseling: mass index (BMI, calculated as the weight

n n
520 Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007
www.mcpdigitalhealth.org
CLINICAL DECISION SUPPORT SYSTEMS IN INFERTILITY

in kilograms divided by the height in me-

TABLE 2. CDSA to Support for Prediction, Decision-Making, and Management of
ters squared), upper limit of menstrual
Patients at Different ART Stages.
cycle length, serum AMH levels, and
CDSA No. of articles identified References
basal androstenedione levels.14
5. A predictive model to estimate outcomes Prediction of treatment success 51 6,8,10,12-59
based on data from pretreatment (ie, based on patient
before starting the first cycle of IVF) characteristics/patient
prognosis
and posttreatment (ie, before starting
the second cycle of IVF in those couples Sperm analysis 11 43,60-69
whose first complete cycle was Decision-making and 9 8,9,32,70-75
unsuccessful).6 counseling to support OS at
various stages
d CDSA for clinical management:
6. A ML model to evaluate gonadotropin Prediction of embryo quality/ 50 22,39,76-123
laboratory management and
starting dose based on candidate’s
embryo selection
characteristics.9,70
7. A ML model to predict the day of oocyte
trigger.8 10. A calculator to estimate blastulation rate
8. An algorithm to estimate the optimal of metaphase II (MII) oocytes and to
number of oocytes to be fertilized based predict the chance of pregnancy.40
on the day of transfer, the number of 11. A calculator of treatment success rates
viable blastocysts obtained, and the based on the patient’s age, the number
number of blastocysts needed to obtain of blastocysts to be transferred in
1 live birth.15 sequence, and the preimplantation diag-
9. A CDSA to manage clinical decisions nosis of aneuploidies.16
regarding the following: (1) continuation All the above-mentioned CDSAs require
of stimulation and (2) in case of discontin- further validation through prospective studies.
uation, whether to trigger or cancel the cy- However, the tools identified may be consid-
cle or (3) in case of continuation of OS, ered in the ART programs given the estab-
whether to determine the days to follow- lished benefit compared with human
up and the need for dosage adjustment.32 decision only.

BMI, Habits, Number of previous

Number of
treatments, Presumptive diagnosis of
COH with FOI calculation oocytes
infertility and estimate of IVF cycle
result
Number of MII
Clinical management

oocytes
Gonadotropins starting
Male Sperm Production &
Embryology
Counseling

dose
Endocrinological profile
Fertilization Rate
Day of oocyte maturation
Female Ovarian Reserve & triggering
Endocrinological profile. Blastulation rate

ET timing & predict

Estimate of oocytes yield, including successuful Implantation Embryo quality
the risk for OHSS evaluation

FIGURE 2. Steps of the infertile couple management journey for which clinical decision support algorithms have been proposed based
on the findings of this review. BMI, body mass index; COH, controlled ovarian hyperstimulation; ET, embryo transfer; FOI, follicle-to-
oocyte index; MII, metaphase II; OHSS, ovarian hyperstimulation syndrome.

n n
Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007 521
www.mcpdigitalhealth.org
 MAYO CLINIC PROCEEDINGS: DIGITAL HEALTH

Automated Tools for ART markers to recommend personalized stra-

The implementation of predictive tools in ART is tegies.10,79,127 AI systems further enhance
indeed expected to yield various positive out- this potential by analyzing extensive data
comes. First, patients can expect more accurate to predict outcomes and continually
prognostic predictions, thus setting more real- refine treatment approaches through ML.
istic expectations. Second, clinicians will benefit 2. CDSA: These systems combine AI and clin-
from streamlined decision-making processes ical expertise to provide real-time decision
leading to optimized treatment strategies and support, helping clinicians to optimize
improved patient outcomes. treatment protocols based on comprehen-
The use of automated technologies in repro- sive patient profiles.4
ductive medicine is increasingly shaping how 3. Predictive calculators: Using mathematical
treatments are tailored and delivered (Table 2, models, these calculators estimate the success
Figure 2). Most of the studies identified evaluated likelihood of ART procedures, factoring in
the efficiency of automated models without direct patient-specific variables such as age and
comparison with human performance BMI. This approach supports clinicians in
(Supplemental Table 1, available online at creating more accurate and personalized treat-
https://www.mcpdigitalhealth.org/), whereas a ment plans.
few made a direct comparison (Supplemental
These technologies not only streamline
Table 2, available online at https://www.
clinical processes but also empower patients
mcpdigitalhealth.org/). Further, we discuss the
by providing clearer insights into their
different tools and their advantages for each step
treatment options, enhancing the transpar-
of the ART journey. For each step, we also
ency and personalization of care in ART.
mention the studies that reported a significant
This shift toward data-driven clinical
advantage of automated tools vs humans.
decision-making will ultimately improve
1. Algorithms and AI in ART: Algorithms the accuracy and effectiveness of infertility
are crucial in ART for making treatment treatments, thereby promising better out-
decisions by integrating various factors comes for patients navigating these chal-
such as age, ovarian reserve, and genetic lenging processes.

TABLE 3. Predictive Calculators for IVF Programs.

Type of cycle Calculator Variables considered Link
Patients using own gamete SART IVF success estimator Age, previous IVF cycles, and https://w3.abdn.ac.uk/clsm/
specific treatment details SARTIVF/tool/ivf1
Boston IVF success rates Age, infertility diagnosis, and https://www.bostonivf.com/
calculator previous IVF outcomes search/?q¼donation
Donor IVF programs SART donor egg calculator Age of the donor, the https://www.sartcorsonline.
recipient’s age, and other com/
treatment variables
FertilityIQ donor egg IVF Age of the donor, the https://www.fertilityiq.com/
success recipient’s age, and the
number of embryos
transferred
Gestational carrier IVF SART gestational carrier IVF Gestational carrier’s age, the https://www.sartcorsonline.
programs calculator intended parent’s age, and com/
treatment specifics
FertilityIQ gestational carrier Gestational carrier’s age, the https://www.fertilityiq.com/
IVF success rate calculator intended parent’s age, and
the number of embryos
transferred
IVF, in vitro fertilization; SART, Society for Assisted Reproductive Technology.

n n
522 Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007
www.mcpdigitalhealth.org
CLINICAL DECISION SUPPORT SYSTEMS IN INFERTILITY

Tools for Patient Prognosis and Counseling AI and Algorithms to Optimize Clinical
The integration of automated technologies in Management
reproductive medicine has the potential to Artificial intelligence has the potential to opti-
enhance treatment outcomes through personal- mize personalized treatment plans by refining
ized protocols and optimization of OS processes: OS protocols and medication dosages, thereby
enhancing ART outcomes. Additionally, wear-
d Algorithms and calculators use patient-
able devices and AI applications facilitate
specific factors to predict treatment out-
continuous patient monitoring and timely
comes effectively. The Pregnancy Probability
interventions.
Calculator from the Institute for Reproduc-
tive Health at Georgetown University uses d Artificial intelligence in OS optimization
variables such as age, BMI, and diagnosis tailor OS protocols by analyzing factors
to predict pregnancy success.128 such as ovarian reserve markers and hor-
d Age-related predictive models estimate the mone levels to optimize the gonadotropin
chances of live birth based on age and other dosage and timing.9,13
factors; for example, the Fertility Potential d Mathematical models for ovarian response
Calculator from the Society for Assisted prediction predict responses based on vari-
Reproductive Technology.129 ables such as AMH levels and antral follicle
d Clinical conditions and diagnostic algo- count, aiding in personalizing treatment
rithms use comprehensive data to diagnose plans and optimizing resource allocation.
conditions like PCOS, using criteria such d Algorithms for cycles personalization iden-
as the Rotterdam criteria.14,17 tify personalized medication dosages during
OS enhance follicular development and
Predictive models in ART show potential in
minimize risks associated with poor or
forecasting pregnancy and live birth rates but
excessive responses.131
require larger, more diverse data sets for
improved accuracy and generalizability. AI Leveraging AI in personalized IVF treat-
models analyze patient data such as age, hormone ment plans can streamline the workflow in
levels, and medical history to predict ART suc- the clinics and offer transparency and realistic
cess, enabling customized treatment plans and expectations for couples. Personalized treat-
realistic expectations for patients. However, these ment plans can not only maximize oocyte
models are often limited by the homogeneity of yield but also minimize the risk of ovarian hy-
the training data. Two studies evaluating the value perstimulation syndrome, and it can enhance
of AI in predicting miscarriages and implantation scheduling precision for IVF cycles. However,
rate reported significant improvement of AI pre- the successful integration of these tools re-
diction vs human prediction.6,130 quires careful clinician oversight to ensure
AI algorithms can predict conditions such as clinical appropriateness and address chal-
endometriosis by analyzing questionnaire re- lenges related to data quality, interpretability,
sponses and various data sources with a 90% pre- and generalizability.
dictive capacity.12 Noninvasive diagnosis of
conditions such as PCOS use AI algorithms analyze
ultrasound images and hormonal profiles with Automated Technologies in Embryo
higher accuracy and earlier than traditional Assessment
methods.14 In embryo selection, AI enhances the accuracy
AI is also used to evaluate sperm parameters of identifying viable embryos by analyzing im-
and DNA fragmentation, crucial for understand- ages and assessing quality and viability,
ing male fertility issues and improving treatment although performance varies across clinics
plans. Tools such as the YO Home Sperm Test and patient populations. Tools such as Early
leverage smartphone technology for quick and Embryo Viability Assessment use AI to
accurate assessments.60,61,125,126 Despite these improve success rates by selecting the most
advancements, ethical concerns and the need viable embryos.
for robust, transparent models remain signifi- d Artificial intelligence in embryo quality
cant challenges. assessment, like those developed by Kragh

n n
Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007 523
www.mcpdigitalhealth.org
 MAYO CLINIC PROCEEDINGS: DIGITAL HEALTH

and Karstoft,132 enhance embryo selection informed decision-making. However, inte-

by analyzing morphologic data to predict grating AI in IVF programs has shown promise
viability, improving accuracy and consis- in enhancing fertility treatments. Clinicians
tency in embryo selection. must educate patients about AI’s role to
d Time-lapse imaging applies mathematical improve transparency, manage expectations,
models to time-lapse data to analyze mor- and prevent dissatisfaction. Transparency en-
phokinetic parameters, aiding in embryo sures patients are aware of the technologies
viability assessment and timing for in their treatment, whereas informed consent
transfer.133,134 requires patients to understand and agree to
d Artificial intelligenceeassisted sperm anal- the use of AI tools, knowing their benefits
ysis enhances the accuracy and efficiency and limitations. Proper education fosters trust
of sperm analysis, which is crucial for select- in both the treatment process and the clini-
ing sperm with the best fertilization cians. Patients should be made aware that AI
potential. effectiveness depends on high-quality data,
and although it can improve success probabil-
Although AI has not yet fully replaced hu-
ities, it does not guarantee positive outcomes.
man expertise in ART laboratories, its integra-
Patients should also be informed about data
tion has significantly enhanced the precision
privacy and ethical considerations. Clinicians
and consistency of embryo quality assess-
should explain how AI is applied at specific
ments. Despite advancements, ongoing model
IVF stages, emphasizing that AI supports,
training and validation are essential owing to
not replaces, fertility specialists’ expertise.
subjective variability in embryo assessment
Therefore, patients should receive compre-
and differing clinic practices. Continued
hensive information about AI use in treatment.
research and development, along with the cre-
Questions should be encouraged and any con-
ation of standardized data sets, are essential for
cerns should be addressed. Providing bro-
maximizing the potential of AI in this field.
chures, videos, personalized sessions, and
group discussions can help patients under-
Automated Technologies for Patient stand AI role in IVF and share experiences, ul-
Empowerment timately building trust and confidence in using
The integration of AI, algorithms, and calcula- advanced technologies in fertility treatments.
tors into reproductive medicine facilitates a
more tailored and efficient approach to treat- Implementation in the Clinics
ment, potentially increasing the likelihood of Although these technologies offer significant
success in ART procedures (Tables 2 and 3). advantages, their implementation in clinical
The main advantages for the patients are as settings requires careful validation to ensure
follows: accuracy and effectiveness. The integration of
AI tools enhances clinical interventions when
d Personalized treatment: Algorithms and cal-
combined with human expertise. The synergy
culators consider individual patient factors
between AI capabilities and clinicians’ efforts
to customize treatment protocols, thereby
leads to precise execution of each intervention
increasing efficacy.
step, transparent follow-up processes, and the
d Improved ovarian response: AI-assisted
development of detailed, strategic plans for
optimization leads to better oocyte yield
achieving cumulative results over time.
and fewer adverse effects.
Currently, AI-based diagnostics for assess-
d Informed decision-making: Treatment
ing embryo quality and predicting pregnancy
outcome calculators provide patients with
outcomes are not comparable with the accu-
personalized prognostic information, aiding
racy of invasive prenatal diagnostics or nonin-
in informed decision-making and opti-
vasive fetal evaluations conducted through
mizing treatment planning.
human ultrasound. Furthermore, the use of
Overall automated systems can improve AI raises concerns, particularly regarding the
transparent communication of predicted treat- need to preserve data privacy. To this purpose,
ment outcomes, thereby providing patients a combination of stringent protocols (ie, Gen-
with realistic expectations and aiding in eral Data Protection Regulation Compliance;
n n
524 Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007
www.mcpdigitalhealth.org
CLINICAL DECISION SUPPORT SYSTEMS IN INFERTILITY

Health Insurance Portability and Account- collection and use practices through clear
ability Act Guidelines; California Consumer privacy policies and consent forms. Users
Privacy Act Regulations; ISO 27001 Stan- are given control over their data, including
dards) and advanced technologies must be options to opt-out of data collection or
applied. The key technologies are as follows: request data deletion.
d Incident response plans: Robust incident
d Data encryption: Data are encrypted both in
response plans are in place to quickly
transit and at rest. This means that any data
address any data breaches or security inci-
being transferred over networks or stored in
dents. These plans include protocols for
databases are converted into a secure code
detecting, reporting, and mitigating the
that can only be decrypted by authorized
impact of data breaches.
users. Encryption standards such as
advanced encryption standard and Secure Using these advanced privacy-preserving
Sockets Layer/Transport Layer Security are techniques and adhering to strict regulatory
commonly used to safeguard sensitive standards, AI tools can ensure the confidenti-
information. ality and security of sensitive data, thereby
d Access controls: Strict access controls are maintaining user trust and adherence with
implemented to ensure that only authorized legal requirements.
personnel can access sensitive data. This in-
cludes multifactor authentication, role-based DISCUSSION
access control, and regularly updated access This systematic review focused on the trans-
permissions. These measures help in formative impact of AI and CDSA for the diag-
limiting data access to only those individuals nosis and treatment of infertility by
who need it for their specific role. significantly enhancing personalized medical
d Anonymization and deidentification: To pro- approaches based on patient predicted out-
tect personal information, AI tools often ano- comes. Table 3 illustrates the predictive capa-
nymize or deidentify data. This process bilities of AI in various stages of ART, from
removes or obfuscates personal identifiers, calculating success rates to evaluating embryo
making it difficult to trace the data back to quality.
an individual. Techniques include data mask- Artificial intelligence technologies use vast
ing, pseudonymization, and aggregation. data sets to analyze variables such as genetic
d Data minimization: AI systems are designed profiles and lifestyle factors, offering insights
to collect only the minimum amount of data for diagnosis and treatment.4,135 Artificial in-
necessary for their function. This principle telligence algorithms exceldwhen adequately
of data minimization reduces the risk of trained on evidence-based pathologiesdin
exposure by limiting the volume of sensitive identifying subtle variations in medical imag-
data being handled. ing.12,91 Artificial intelligence technologies
d Secure data storage: Data are stored in secure are also a valuable tool for treatment
environments with robust physical and decision-making. They assist in optimizing
cybersecurity measures. This includes secure infertility treatment protocols by analyzing
servers, data centers with restricted access, patient-specific factors such as BMI and hor-
and cloud storage solutions that comply mone levels, thus personalizing medication
with industry standards such as ISO 27001. dosages and timing.10 Moreover, AI enhances
d Regular audits and adherence: Regular secu- embryo assessment by analyzing morphoki-
rity audits and adherence checks are conduct- netic and genetic data to select embryos with
ed to ensure adherence to data protection the highest potential for successful preg-
regulations such as General Data Protection nancy.136 Artificial intelligence’s objective
Regulation, Health Insurance Portability and analysis helps in reducing human error in em-
Accountability Act, and California Consumer bryo selection, thereby improving the chances
Privacy Act. These audits help identify and of successful implantations.
mitigate potential vulnerabilities. Artificial intelligence and CDSA have the
d User consent and transparency: AI tools potential to provide personalized treatment
ensure that users are informed about data plans that potentially reduce the number of

n n
Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007 525
www.mcpdigitalhealth.org
 MAYO CLINIC PROCEEDINGS: DIGITAL HEALTH

treatment cycles required, lessening the Notably, the Federal Drug Administration
emotional and financial burden on couples. AI now regulates CDSA systems as medical de-
is based on learning from the training sets, vices, emphasizing the importance of reliable
which is subjective to variability, and that the predictions in clinical decision-making.5 The
outcome is simply bound to the different vari- use of AI and CDSA will not replace physi-
ables examined in the testing process. cians and embryologists, but it will help
Furthermore, AI is expected to improve im- enhance efficiency and quality of work,
aging technologies, integrate genomic data for thereby increasing access to care, reducing
better predictions, and facilitate remote ART ser- costs, and waiting times.137
vices through telemedicine, enhancing patient
accessibility and treatment outcomes. Ethical
considerations, including ensuring data privacy
CONCLUSION
and obtaining patient consent, are crucial for
In conclusion, the advent of automated tech-
responsible AI use in clinical settings.
nologies is transforming the field of infertility
Artificial intelligence is significantly
diagnosis and treatment. Algorithms have the
advancing the field of reproductive medicine
potential to expedite and enhance the accuracy
by providing more accurate diagnoses and
of diagnosis, guide personalized treatment,
personalized treatment options. Although AI
and improve overall patient care. As in some
and ML models offer promising advancements,
specific medical applications, AI continues to
a critical review of their efficacy, limitations, and
evolve.138,139 In the near future, we can expect
overall impact is necessary to understand their
further advances that will bring new hope to
real-world applicability and potential for
couples struggling with infertility, to help
improving ART outcomes. In particular, studies
achieve their dream of starting a family.140
evaluating the application of AI should consider
Indeed, automated tools may enable re-
the following critical aspects:
searchers and clinicians to gain deeper insights
d Data quality and quantity: The effectiveness into the complex dynamics of fertility. By inte-
of AI/ML models depends heavily on the grating diverse data sets and simulating
quality and quantity of data available. Incon- various scenarios, these models promise to
sistent or incomplete data can lead to inac- enhance diagnostic penetrance, optimize treat-
curate predictions. ment strategies, and improve treatment out-
d Generalizability: Models trained on specific comes for infertile couples.
data sets may not perform well when Overall, the validated use of AI, mathemat-
applied to different populations or clinical ical models, algorithms, and calculators in the
settings. This raises concerns about the clinical management of infertile couples dur-
generalizability of the results. ing OS yields significant advantages. These
d Interpretability: Many AI/ML models, partic- tools prove invaluable not only in subsequent
ularly deep learning models, operate as steps of IVF procedures but also during labo-
“black boxes,” making it difficult to under- ratory/embryology management and embryo
stand how decisions are made. This lack of transfer. However, further research and valida-
transparency can be a barrier to clinical tion studies are essential to continually refine
acceptance and patient trust. and broaden the applications of these tools
d Ethical and legal concerns: The use of patient in both clinical and laboratory settings.
data in AI/ML models raises ethical and legal Indeed, at the current stage, it is not possible
questions about privacy, consent, and data se- to draw any conclusions from the results of
curity. Depending on specific country body this review owing to the lack of standardiza-
laws different countries may interpretate the tion in validation methods across studies.
AI/ML integration in the clinical practice as Further validation and clinical trials are
issue to persecute or value to encourage. How- needed to establish the effectiveness of these
ever, the use of AI/ML remains the doctor’s legal tools in the clinical setting.
responsibility. In fact, in several countries, it is With ongoing advancements, the integra-
not possible to transfer that responsibility tion of AI, mathematical models, algorithms,
from doctor to AI/ML tools. and calculators promises to further enhance

n n
526 Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007
www.mcpdigitalhealth.org
CLINICAL DECISION SUPPORT SYSTEMS IN INFERTILITY

the precision and efficacy of medical and Department of Obstetrics and Gynecology, Stanford Uni-
embryologic protocols, thereby leading to versity School of Medicine, Sunnyvale, CA (L.A.); Depart-
ment Obstetrics and Gynecology, University Hospital of
improved treatment outcomes and heightened
Vaud, Lausanne, Switzerland (F.M.B.); ART Fertility Clinics,
success rates in ART. In the foreseeable future, Dubai, United Arab Emirates (B.A.); and Department of Ob-
a superalgorithm that integrates all computer stetrics and Gynecology, Koç University School of Medicine,
and AI tools that have been clinically validated Istanbul, Turkey (B.A.).
could be adopted in the ART clinics by fertility
experts and embryologists, leading to signifi-
Grant Support: This work was supported by unrestricted
cant benefits for all patients involved. grant from IBSA. The funder of the study had no role in
study design, data collection, data analysis, data interpreta-
POTENTIAL COMPETING INTERESTS tion, or writing of the report.
The authors report no competing interests.
Correspondence: Address to Carlo Bulletti, MD, Help Me
Doctor, Via Nazario Sauro 30, 47841 Cattolica (RN), Italy
DECLARATION OF GENERATIVE AI AND AI- ([email protected]).
ASSISTED TECHNOLOGIES IN THE WRITING
PROCESS
The authors employed AI and AI-assisted REFERENCES
technologies to enhance the readability and 1. European IVF Monitoring Consortium (EIM) for the European So-
language of this work. These tools were used ciety of Human Reproduction and Embryology (ESHRE), Smeenk J,
Wyns C, Geyter De, et al. ART in Europe, 2019: results generated
to support, not replace, essential authoring
from European registries by ESHRE. Hum Reprod. 2023;38(12):
tasks such as generating scientific, pedagogic, 2321-2338. https://doi.org/10.1093/humrep/dead197.
or medical insights, drawing scientific conclu- 2. Aimagambetova G, Issanov A, Terzic S, et al. The effect of
sions, and providing clinical recommenda- psychological distress on IVF outcomes: reality or specula-
tions? PLoS One. 2020;15(12):e0242024. https://doi.org/10.
tions. Human oversight and control were 1371/journal.pone.0242024.
maintained throughout the writing process, 3. Ben Rafael Z. Repeated implantation failure (RIF): an iatro-
genic meaningless definition that generates unnecessary and
and all work was carefully reviewed and edi-
costly use of add-on procedures. Hum Reprod. 2020;35(7):
ted. The authors remain fully responsible 1479-1483. https://doi.org/10.1093/humrep/deaa134.
and accountable for the content of this work. 4. Hanassab S, Abbara A, Yeung AC, et al. The prospect of arti-
ficial intelligence to personalize assisted reproductive technol-
ogy. NPJ Digit Med. 2024;7(1):55. https://doi.org/10.1038/
ACKNOWLEDGMENT s41746-024-01006-x.
5. Goodman KE, Rodman AM, Morgan DJ. Preparing physicians
The authors thank Daniela Criscuolo for the
for the clinical algorithm era. N Engl J Med. 2023;389(6):483-
systematic review and Daniela Finizio and 487. https://doi.org/10.1056/NEJMp2304839.
Roberta Fabiani (Scientific Communication 6. McLernon DJ, Raja E-A, Toner JP, et al. Predicting personalized cu-
mulative live birth following in vitro fertilization. Fertil Steril. 2022;
srl, Naples, Italy) for supporting the systematic
117(2):326-338. https://doi.org/10.1016/j.fertnstert.2021.09.015.
review and editorial coordination. 7. Thirunavukarasu AJ, Ting DSJ, Elangovan K, Gutierrez L,
Tan TF, Ting DSW. Large language models in medicine. Nat
Med. 2023;29(8):1930-1940. https://doi.org/10.1038/s41
SUPPLEMENTAL ONLINE MATERIAL 591-023-02448-8.
Supplemental material can be found online at 8. Fanton M, Nutting V, Solano F, et al. An interpretable machine
learning model for predicting the optimal day of trigger during
https://www.mcpdigitalhealth.org/. Supple-
ovarian stimulation. Fertil Steril. 2022;118(1):101-108. https://
mental material attached to journal articles doi.org/10.1016/j.fertnstert.2022.04.003.
has not been edited, and the authors take re- 9. Fanton M, Nutting V, Rothman A, et al. An interpretable machine
learning model for individualized gonadotrophin starting dose se-
sponsibility for the accuracy of all data.
lection during ovarian stimulation. Reprod Biomed Online. 2022;
45(6):1152-1159. https://doi.org/10.1016/j.rbmo.2022.07.010.
Abbreviations and Acronyms: AI, artificial intelligence; 10. Wang C-W, Kuo C-Y, Chen C-H, Hsieh Y-H, Su EC-Y. Predict-
AMH, anti-Müllerian hormone; ART, assisted reproduction ing clinical pregnancy using clinical features and machine learning
algorithms in in vitro fertilization. PLoS One. 2022;17(6):
technology; CDSA, clinical decision support algorithm; IVF,
e0267554. https://doi.org/10.1371/journal.pone.0267554.
in vitro fertilization; ML, machine learning; OS, ovarian 11. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group.
stimulation; PCOS, polycystic ovary syndrome Preferred reporting items for systematic reviews and meta-
analyses: the PRISMA statement. PLoS Med. 2009;6(7):
Affiliations (Continued from the first page of this e1000097. https://doi.org/10.1371/journal.pmed.1000097.
article.): Woman Mother Child, Lausanne University Hospi- 12. Chapron C, Lafay-Pillet M-C, Santulli P, et al. A new validated
tal, CHUV, Lausanne, Switzerland (R.S.); Department of Sta- screening method for endometriosis diagnosis based on patient
tistical Sciences, University of Bologna, Bologna, Italy (M.B.); questionnaires. EclinicalMedicine. 2022;44:101263. https://doi.
Division of Reproductive Endocrinology and Infertility, org/10.1016/j.eclinm.2021.101263.

n n
Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007 527
www.mcpdigitalhealth.org
 MAYO CLINIC PROCEEDINGS: DIGITAL HEALTH

13. Esteves SC, Roque M, Sunkara SK, et al. Oocyte quantity, as well 29. Barnett-Itzhaki Z, Elbaz M, Butterman R, et al. Machine
as oocyte quality, plays a significant role for the cumulative live learning vs. classic statistics for the prediction of IVF outcomes.
birth rate of a POSEIDON criteria patient. Hum Reprod. 2019; J Assist Reprod Genet. 2020;37(10):2405-2412. https://doi.org/
34(12):2555-2557. https://doi.org/10.1093/humrep/dez181. 10.1007/s10815-020-01908-1.
14. Xu H, Feng G, Shi L, Han Y, Huang Q, Li R. PCOSt: A non- 30. Ferrick L, Lee YSL, Gardner DK. Metabolic activity of human blasto-
invasive and cost-effective screening tool for polycystic ovary cysts correlates with their morphokinetics, morphological grade,
syndrome. Innovation (Camb). 2023;4(2):100407. https://doi. KIDScore and artificial intelligence ranking. Hum Reprod. 2020;
org/10.1016/j.xinn.2023.100407. 35(9):2004-2016. https://doi.org/10.1093/humrep/deaa181.
15. Correia KFB, Missmer SA, Weinerman R, Ginsburg ES, Rossi BV. 31. Goyal A, Kuchana M, Ayyagari KPR. Machine learning predicts live-
Development of a model to estimate the optimal number of oo- birth occurrence before in-vitro fertilization treatment. Sci Rep.
cytes to attempt to fertilize during assisted reproductive technol- 2020;10(1):20925. https://doi.org/10.1038/s41598-020-76928-z.
ogy treatment. JAMA Netw Open. 2023;6(1):e2249395. https:// 32. Letterie G, MacDonald A. Artificial intelligence in in vitro fertil-
doi.org/10.1001/jamanetworkopen.2022.49395. ization: a computer decision support system for day-to-day
16. Ata B, Kalafat E, Somigliana E. A new definition of recurrent management of ovarian stimulation during in vitro fertilization.
implantation failure on the basis of anticipated blastocyst Fertil Steril. 2020;114(5):1026-1031. https://doi.org/10.1016/j.
aneuploidy rates across female age. Fertil Steril. 2021;116(5): fertnstert.2020.06.006.
1320-1327. https://doi.org/10.1016/j.fertnstert.2021.06.045. 33. Liao S, Pan W, Dai W-Q, et al. Development of a dynamic
17. Zad Z, Jiang VS, Wolf AT, et al. Predicting polycystic ovary diagnosis grading system for infertility using machine learning.
syndrome with machine learning algorithms from electronic JAMA Netw Open. 2020;3(11):e2023654. https://doi.org/10.
health records. Front Endocrinol (Lausanne). 2024;15: 1001/jamanetworkopen.2020.23654.
1298628. https://doi.org/10.3389/fendo.2024.1298628. 34. Liu L, Jiao Y, Li X, Ouyang Y, Shi D. Machine learning algo-
18. Kaufmann SJ, Eastaugh JL, Snowden S, Smye SW, Sharma V. rithms to predict early pregnancy loss after in vitro
The application of neural networks in predicting the outcome fertilization-embryo transfer with fetal heart rate as a strong
of in- vitro fertilization. Hum Reprod. 1997;12(7):1454-1457. predictor. Comput Methods Programs Biomed. 2020;196:
https://doi.org/10.1093/humrep/12.7.1454. 105624. https://doi.org/10.1016/j.cmpb.2020.105624.
19. Wald M, Sparks AET, Sandlow J, Van-Voorhis B, Syrop CH, 35. Liu R, Bai S, Jiang X, et al. Multifactor prediction of embryo
Niederberger CS. Computational models for prediction of transfer outcomes based on a machine learning algorithm.
IVF/ICSI outcomes with surgically retrieved spermatozoa. Front Endocrinol (Lausanne). 2021;12:745039. https://doi.org/
Reprod Biomed Online. 2005;11(3):325-331. https://doi.org/ 10.3389/fendo.2021.745039.
10.1016/s1472-6483(10)60840-1. 36. Liu H, Zhang Z, Gu Y, et al. Development and evaluation of a
20. Jones CA, Christensen AL, Salihu H, et al. Prediction of indi- live birth prediction model for evaluating human blastocysts
vidual probabilities of livebirth and multiple birth events from a retrospective study. Elife. 2023;12:e83662. https://doi.
following in vitro fertilization (IVF): a new outcomes counsel- org/10.7554/eLife.83662.
ling tool for IVF providers and patients using HFEA metrics. 37. Raef B, Maleki M, Ferdousi R. Computational prediction of implan-
J Exp Clin Assist Reprod. 2011;8:3. tation outcome after embryo transfer. Health Informatics J. 2020;
21. Uyar A, Bener A, Ciray HN. Predictive modeling of implanta- 26(3):1810-1826. https://doi.org/10.1177/1460458219892138.
tion outcome in an in vitro fertilization setting: an application 38. Tarín JJ, Pascual E, García-Pérez MA, Gómez R, Hidalgo-
of machine learning methods. Med Decis Making. 2015;35(6): Mora JJ, Cano A. A predictive model for women’s assisted
714-725. https://doi.org/10.1177/0272989X14535984. fecundity before starting the first IVF/ICSI treatment cycle.
22. Milewski R, Kuczy
nska A, Stankiewicz B, Kuczy
nski W. How J Assist Reprod Genet. 2020;37(1):171-180. https://doi.org/10.
much information about embryo implantation potential is 1007/s10815-019-01642-3.
included in morphokinetic data? A prediction model based 39. Huang B, Tan W, Li Z, Jin L. An artificial intelligence model
on artificial neural networks and principal component analysis. (euploid prediction algorithm) can predict embryo ploidy sta-
Adv Med Sci. 2017;62(1):202-206. https://doi.org/10.1016/j. tus based on time-lapse data. Reprod Biol Endocrinol. 2021;
advms.2017.02.001. 19(1):185. https://doi.org/10.1186/s12958-021-00864-4.
23. Burai P, Hajdu A, Manuel FE, Harangi B. Segmentation of the 40. Jin H, Shen X, Song W, Liu Y, Qi L, Zhang F. The develop-
uterine wall by an ensemble of fully convolutional neural net- ment of nomograms to predict blastulation rate following cy-
works. Annu Int Conf IEEE Eng Med Biol Soc. 2018;2018:49-52. cles of in vitro fertilization in patients with tubal factor
https://doi.org/10.1109/EMBC.2018.8512245. infertility, polycystic ovary syndrome, or endometriosis. Front
24. Leijdekkers JA, Eijkemans MJC, van Tilborg TC, et al. Predicting Endocrinol (Lausanne). 2021;12:751373. https://doi.org/10.
the cumulative chance of live birth over multiple complete cy- 3389/fendo.2021.751373.
cles of in vitro fertilization: an external validation study. Hum 41. Kozar N, Kovac V, Reljic M. Can methods of artificial intelligence
Reprod. 2018;33(9):1684-1695. https://doi.org/10.1093/ aid in optimizing patient selection in patients undergoing intra-
humrep/dey263. uterine inseminations? J Assist Reprod Genet. 2021;38(7):1665-
25. Blank C, Wildeboer RR, DeCroo I, et al. Prediction of implan- 1673. https://doi.org/10.1007/s10815-021-02224-y.
tation after blastocyst transfer in in vitro fertilization: a 42. Li F, Lu R, Zeng C, Li X, Xue Q. Development and validation of a
machine-learning perspective. Fertil Steril. 2019;111(2):318- clinical pregnancy failure prediction model for poor ovarian re-
326. https://doi.org/10.1016/j.fertnstert.2018.10.030. sponders during IVF/ICSI. Front Endocrinol (Lausanne). 2021;12:
26. Qiu J, Li P, Dong M, Xin X, Tan J. Personalized prediction of 717288. https://doi.org/10.3389/fendo.2021.717288.
live birth prior to the first in vitro fertilization treatment: a ma- 43. Srivastava D, Gupta S, Kudavelly S, Suryanarayana KV, Ga R.
chine learning method. J Transl Med. 2019;17(1):317. https:// Unsupervised deep learning based longitudinal follicular
doi.org/10.1186/s12967-019-2062-5. growth tracking during IVF cycle using 3D transvaginal ultra-
27. Vogiatzi P, Pouliakis A, Siristatidis C. An artificial neural sound in assisted reproduction. Annu Int Conf IEEE Eng Med
network for the prediction of assisted reproduction outcome. Biol Soc. 2021;2021:3209-3212.
J Assist Reprod Genet. 2019;36(7):1441-1448. https://doi.org/ 44. Bardet L, Excoffier J-B, Salaun-Penquer N, et al. Comparison
10.1007/s10815-019-01498-7. of predictive models for cumulative live birth rate after treat-
28. Xu B, Liu C, Qian L, et al. Statistical modelling outcome of in vitro ment with ART. Reprod Biomed Online. 2022;45(2):246-255.
fertilization and intracytoplasmic sperm injection: a single centre https://doi.org/10.1016/j.rbmo.2022.03.020.
study. Comb Chem High Throughput Screen. 2019;22(4):225- 45. Fu K, Li Y, Lv H, Wu W, Song J, Xu J. Development of a model
231. https://doi.org/10.2174/1386207322666190404145448. predicting the outcome of in vitro fertilization cycles by a

n n
528 Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007
www.mcpdigitalhealth.org
CLINICAL DECISION SUPPORT SYSTEMS IN INFERTILITY

robust decision tree method. Front Endocrinol. 2022;13: motility for artificial insemination. Sensors (Basel). 2020;21(1):
877518. https://doi.org/10.3389/fendo.2022.877518. 72. https://doi.org/10.3390/s21010072.
46. Kuroda S, Karna KK, Kaiyal RS, Gupta S, Sharma R, Agarwal A. Devel- 62. Javadi S, Mirroshandel SA. A novel deep learning method for
opment of a novel robust artificial intelligence developed sperm automatic assessment of human sperm images. Comput Biol
DNA fragmentation testdpreliminary findings. Fertil Steril. 2022; Med. 2019;109:182-194. https://doi.org/10.1016/j.compbiomed.
118(4):e307. https://doi.org/10.1016/j.fertnstert.2022.09.064. 2019.04.030.
47. Mehrjerd A, Rezaei H, Eslami S, Ratna MB, Khadem Ghaebi N. 63. McCallum C, Riordon J, Wang Y, et al. Deep learning-based
Internal validation and comparison of predictive models to selection of human sperm with high DNA integrity. Commun
determine success rate of infertility treatments: a retrospective Biol. 2019;2:250. https://doi.org/10.1038/s42003-019-0491-6.
study of 2485 cycles. Sci Rep. 2022;12(1):7216. https://doi.org/ 64. Abbasi A, Miahi E, Mirroshandel SA. Effect of deep transfer
10.1038/s41598-022-10902-9. and multi-task learning on sperm abnormality detection. Com-
48. Mehrjerd A, Rezaei H, Eslami S, Khadem Ghaebi N. Determi- put Biol Med. 2021;128:104121. https://doi.org/10.1016/j.
nation of cut off for endometrial thickness in couples with un- compbiomed.2020.104121.
explained infertility: trustable AI. Stud Health Technol Inform. 65. Jiang A, Jiaqi W, Zhao H, Zhang Z, Sun Y. Identifying viability
2022;294:264-268. https://doi.org/10.3233/SHTI220450. of immotile sperm at one glance: sperm viability classifier
49. Mercuri N, Fjeldstad J, Krivoi A, Meriano J, Nayot D. A non-inva- powered by deep learning. Fertil Steril. 2022;118:e297-e298.
sive, 2-dimensional (2D) image analysis artificial intelligence (AI) https://doi.org/10.1016/j.fertnstert.2022.08.828.
tool scores mature oocytes and correlates with the quality of 66. Mendizabal-Ruiz G, Chavez-Badiola A, Aguilar Figueroa I, et al.
subsequent blastocyst development. Fertil Steril. 2022;118(4): Computer software (SiD) assisted real-time single sperm se-
e78-e79. https://doi.org/10.1016/j.fertnstert.2022.08.241. lection associated with fertilization and blastocyst formation.
50. Qu P, Chen L, Zhao D, Shi W, Shi J. Nomogram for the cu- Reprod Biomed Online. 2022;45:703-711. https://doi.org/10.
mulative live birth in women undergoing the first IVF cycle: 1016/j.rbmo.2022.03.036.
Base on 26, 689 patients in China. Front Endocrinol (Lausanne). 67. Ottl S, Amiriparian S, Gerczuk M, Schuller BW. motilitAI: a
2022;13:900829. https://doi.org/10.3389/fendo.2022.900829. machine learning framework for automatic prediction of hu-
51. Shen L, Zhang Y, Chen W, Yin X. The application of artificial man sperm motility. iScience. 2022;25(8):104644. https://doi.
intelligence in predicting embryo transfer outcome of recur- org/10.1016/j.isci.2022.104644.
rent implantation failure. Front Physiol. 2022;13:885661. 68. Saiffe Farías AF, Sakkas D, Chavez-Badiola A, et al. Single-
https://doi.org/10.3389/fphys.2022.885661. sperm motility analysis during icsi using an artificial intelligence
52. Wen J-Y, Liu C-F, Chung M-T, Tsai Y-C. Artificial intelligence sperm identification software (SID) and correlation with
model to predict pregnancy and multiple pregnancy risk morphology. Fertil Steril. 2022;118(4):e56-e57. https://doi.
following in vitro fertilization-embryo transfer (IVF-ET). Taiwan org/10.1016/j.fertnstert.2022.08.178.
J Obstet Gynecol. 2022;61(5):837-846. https://doi.org/10.1016/ 69. Peng T, Liao C, Ye X, et al. Machine learning-based clustering
j.tjog.2021.11.038. to identify the combined effect of the DNA fragmentation in-
53. Yang L, Peavey M, Kaskar K, et al. Development of a dynamic dex and conventional semen parameters on in vitro fertiliza-
machine learning algorithm to predict clinical pregnancy and tion outcomes. Reprod Biol Endocrinol. 2023;21(1):26. https://
live birth rate with embryo morphokinetics. F S Rep. 2022; doi.org/10.1186/s12958-023-01080-y.
3(2):116-123. https://doi.org/10.1016/j.xfre.2022.04.004. 70. Kobanawa M. The gonadotropins starting dose calculator,
54. Cheredath A, Uppangala S, Asha CS, et al. Combining ma- which can be adjusted the target number of oocytes and
chine learning with metabolomic and embryologic data im- stimulation duration days to achieve individualized
proves embryo implantation prediction. Reprod Sci. 2023; controlled ovarian stimulation in Japanese patients. Reprod
30(3):984-994. https://doi.org/10.1007/s43032-022-01071-1. Med Biol. 2023;22(1):e12499. https://doi.org/10.1002/rmb2.
55. Gong X, Zhang Y, Zhu Y, et al. Development and validation of a 12499.
live birth prediction model for expected poor ovarian response 71. Howles CM, Saunders H, Alam V, Engrand P; FSH Treatment
patients during IVF/ICSI. Front Endocrinol (Lausanne). 2023;14: Guidelines Clinical Panel. Predictive factors and a correspond-
1027805. https://doi.org/10.3389/fendo.2023.1027805. ing treatment algorithm for controlled ovarian stimulation in
56. Louis CM, Handayani N, Aprilliana T, Polim AA, Boediono A, patients treated with recombinant human follicle stimulating
Sini I. Genetic algorithm-assisted machine learning for clinical hormone (follitropin alfa) during assisted reproduction tech-
pregnancy prediction in in vitro fertilization. AJOG Glob Rep. nology (ART) procedures. An analysis of 1378 patients. Curr
2023;3(1):100133. https://doi.org/10.1016/j.xagr.2022.100133. Med Res Opin. 2006;22(5):907-918. https://doi.org/10.1185/
57. Ozer G, Akca A, Yuksel B, Duzguner I, Pehlivanli AC, Kahraman S. 030079906X104678.
Prediction of risk factors for first trimester pregnancy loss in 72. Olivennes F, Trew G, Borini A, et al. Randomized, controlled,
frozen-thawed good-quality embryo transfer cycles using ma- open-label, non-inferiority study of the CONSORT algorithm
chine learning algorithms. J Assist Reprod Genet. 2023;40(2):279- for individualized dosing of follitropin alfa. Reprod Biomed On-
288. https://doi.org/10.1007/s10815-022-02645-3. line. 2015;30(3):248-257. https://doi.org/10.1016/j.rbmo.2014.
58. Yuan G, Lv B, Du X, et al. Prediction model for missed abor- 11.013.
tion of patients treated with IVF-ET based on XGBoost: a 73. Hariton E, Chi EA, Chi G, et al. A machine learning algorithm
retrospective study. PeerJ. 2023;11:e14762. https://doi.org/10. can optimize the day of trigger to improve in vitro fertilization
7717/peerj.14762. outcomes. Fertil Steril. 2021;116(5):1227-1235. https://doi.org/
59. Canosa S, Licheri N, Bergandi L, et al. A novel machine- 10.1016/j.fertnstert.2021.06.018.
learning framework based on early embryo morphokinetics 74. Correa N, Cerquides J, Arcos JL, Vassena R. Supporting first
identifies a feature signature associated with blastocyst devel- FSH dosage for ovarian stimulation with machine learning.
opment. J Ovarian Res. 2024;17(1):63. https://doi.org/10.1186/ Reprod Biomed Online. 2022;45(5):1039-1045. https://doi.org/
s13048-024-01376-6. 10.1016/j.rbmo.2022.06.010.
60. Yi WJ, Park KS, Paick JS. Morphological classification of sperm 75. Letterie G, MacDonald A, Shi Z. An artificial intelligence plat-
heads using artificial neural networks. Stud Health Technol form to optimize workflow during ovarian stimulation and IVF:
Inform. 1998;52(Pt 2):1071-1074. process improvement and outcome-based predictions.
61. Valiuskaite_ V, Raudonis V, Maskeliunas R, Damasevicius R, Reprod Biomed Online. 2022;44(2):254-260. https://doi.org/
Krilavicius T. Deep learning based evaluation of spermatozoid 10.1016/j.rbmo.2021.10.006.

n n
Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007 529
www.mcpdigitalhealth.org
 MAYO CLINIC PROCEEDINGS: DIGITAL HEALTH

76. Morales DA, Bengoetxea E, Larrañaga P, et al. Bayesian classifica- 91. Kovacs P, Matyas S, Forgacs V, Sajgo A, Molnar L,
tion for the selection of in vitro human embryos using morpho- Pribenszky C. Non-invasive embryo evaluation and selection
logical and clinical data. Comput Methods Programs Biomed. 2008; using time-lapse monitoring: results of a randomized
90(2):104-116. https://doi.org/10.1016/j.cmpb.2007.11.018. controlled study. Eur J Obstet Gynecol Reprod Biol. 2019;233:
77. Manna C, Nanni L, Lumini A, Pappalardo S. Artificial intelli- 58-63. https://doi.org/10.1016/j.ejogrb.2018.12.011.
gence techniques for embryo and oocyte classification. Reprod 92. Kragh MF, Rimestad J, Berntsen J, Karstoft H. Automatic grading of
Biomed Online. 2013;26(1):42-49. https://doi.org/10.1016/j. human blastocysts from time-lapse imaging. Comput Biol Med.
rbmo.2012.09.015. 2019;115:103494. https://doi.org/10.1016/j.compbiomed.2019.
78. VerMilyea MD, Tan L, Anthony JT, et al. Computer-auto- 103494.
mated time-lapse analysis results correlate with embryo im- 93. Liu Y, Feenan K, Chapple V, Matson P. Assessing efficacy of
plantation and clinical pregnancy: a blinded, multi-centre day 3 embryo time-lapse algorithms retrospectively: impacts
study. Reprod Biomed Online. 2014;29(6):729-736. https://doi. of dataset type and confounding factors. Hum Fertil (Camb).
org/10.1016/j.rbmo.2014.09.005. 2019;22(3):182-190. https://doi.org/10.1080/14647273.2018.
79. VerMilyea M, Hall JMM, Diakiw SM, et al. Development of an 1425919.
artificial intelligence-based assessment model for prediction of 94. Reignier A, Girard J-M, Lammers J, et al. Performance of Day 5
embryo viability using static images captured by optical light KIDScoreTM morphokinetic prediction models of implantation
microscopy during IVF. Hum Reprod. 2020;35(4):770-784. and live birth after single blastocyst transfer. J Assist Reprod
https://doi.org/10.1093/humrep/deaa013. Genet. 2019;36(11):2279-2285. https://doi.org/10.1007/s1081
80. Diamond MP, Suraj V, Behnke EJ, et al. Using the Eeva TestTM 5-019-01567-x.
adjunctively to traditional day 3 morphology is informative for 95. Tran D, Cooke S, Illingworth PJ, Gardner DK. Deep learning as
consistent embryo assessment within a panel of embryologists a predictive tool for fetal heart pregnancy following time-lapse
with diverse experience. J Assist Reprod Genet. 2015;32(1):61- incubation and blastocyst transfer. Hum Reprod. 2019;34(6):
68. https://doi.org/10.1007/s10815-014-0366-1. 1011-1018. https://doi.org/10.1093/humrep/dez064.
81. Petersen BM, Boel M, Montag M, Gardner DK. Development 96. Bori L, Paya E, Alegre L, et al. Novel and conventional embryo
of a generally applicable morphokinetic algorithm capable of parameters as input data for artificial neural networks: an arti-
predicting the implantation potential of embryos transferred ficial intelligence model applied for prediction of the implanta-
on Day 3. Hum Reprod. 2016;31(10):2231-2244. https://doi. tion potential. Fertil Steril. 2020;114(6):1232-1241. https://doi.
org/10.1093/humrep/dew188. org/10.1016/j.fertnstert.2020.08.023.
82. Adolfsson E, Porath S, Andershed AN. External validation of a 97. Chavez-Badiola A, Flores-Saiffe-Farías A, Mendizabal-Ruiz G,
time-lapse model: a retrospective study comparing embryo eval- Drakeley AJ, Cohen J. Embryo Ranking Intelligent Classification Al-
uation using a morphokinetic model to standard morphology gorithm (ERICA): artificial intelligence clinical assistant predicting
with live birth as endpoint. JBRA Assist Reprod. 2018;22(3):205- embryo ploidy and implantation. Reprod Biomed Online. 2020;
214. https://doi.org/10.5935/1518-0557.20180041. 41(4):585-593. https://doi.org/10.1016/j.rbmo.2020.07.003.
83. Bodri D, Milewski R, Yao Serna J, et al. Predicting live birth by 98. Feyeux M, Reignier A, Mocaer M, et al. Development of auto-
combining cleavage and blastocyst-stage time-lapse variables mated annotation software for human embryo morphoki-
using a hierarchical and a data mining-based statistical model. netics. Hum Reprod. 2020;35(3):557-564. https://doi.org/10.
Reprod Biol. 2018;18(4):355-360. https://doi.org/10.1016/j. 1093/humrep/deaa001.
repbio.2018.10.006. 99. Fishel S, Campbell A, Foad F, et al. Evolution of embryo selec-
84. Rocafort E, Enciso M, Leza A, Sarasa J, Aizpurua J. Euploid em- tion for IVF from subjective morphology assessment to objec-
bryos selected by an automated time-lapse system have supe- tive time-lapse algorithms improves chance of live birth.
rior SET outcomes than selected solely by conventional Reprod Biomed Online. 2020;40(1):61-70. https://doi.org/10.
morphology assessment. J Assist Reprod Genet. 2018;35(9): 1016/j.rbmo.2019.10.005.
1573-1583. https://doi.org/10.1007/s10815-018-1265-7. 100. Fukunaga N, Sanami S, Kitasaka H, et al. Development of an
85. Storr A, Venetis C, Cooke S, Kilani S, Ledger W. Time-lapse al- automated two pronuclei detection system on time-lapse em-
gorithms and morphological selection of day-5 embryos for bryo images using deep learning techniques. Reprod Med Biol.
transfer: a preclinical validation study. Fertil Steril. 2018;109(2): 2020;19(3):286-294. https://doi.org/10.1002/rmb2.12331.
276-283.e3. https://doi.org/10.1016/j.fertnstert.2017.10.036. 101. Blais I, Koifman M, Feferkorn I, Dirnfeld M, Lahav-Baratz S.
86. Strouthopoulos C, Anifandis G. An automated blastomere Improving embryo selection by the development of a
identification method for the evaluation of day 2 embryos laboratory-adapted time-lapse model. F S Sci. 2021;2(2):176-
during IVF/ICSI treatments. Comput Methods Programs Biomed. 197. https://doi.org/10.1016/j.xfss.2021.02.001.
2018;156:53-59. https://doi.org/10.1016/j.cmpb.2017.12.022. 102. Coticchio G, Fiorentino G, Nicora G, et al. Cytoplasmic
87. Alegre L, Del Gallego R, Arrones S, Hernández P, Muñoz M, movements of the early human embryo: imaging and artificial
Meseguer M. Novel noninvasive embryo selection algorithm intelligence to predict blastocyst development. Reprod Biomed
combining time-lapse morphokinetics and oxidative status of Online. 2021;42(3):521-528. https://doi.org/10.1016/j.rbmo.
the spent embryo culture medium. Fertil Steril. 2019;111(5): 2020.12.008.
918-927.e3. https://doi.org/10.1016/j.fertnstert.2019.01.022. 103. Fitz VW, Kanakasabapathy MK, Thirumalaraju P, et al. Should
88. Dirvanauskas D, Maskeliunas R, Raudonis V, Damasevicius R. there be an ‘AI’ in TEAM? Embryologists selection of high im-
Embryo development stage prediction algorithm for automated plantation potential embryos improves with the aid of an arti-
time lapse incubators. Comput Methods Programs Biomed. 2019; ficial intelligence algorithm. J Assist Reprod Genet. 2021;38(10):
177:161-174. https://doi.org/10.1016/j.cmpb.2019.05.027. 2663-2670. https://doi.org/10.1007/s10815-021-02318-7.
89. Kanakasabapathy MK, Thirumalaraju P, Bormann CL, et al. 104. Friedenthal J, Hernandez-Nieto C, Roth RM, et al. Clinical imple-
Development and evaluation of inexpensive automated deep mentation of algorithm-based embryo selection is associated
learning-based imaging systems for embryology. Lab Chip. with improved pregnancy outcomes in single vitrified warmed
2019;19(24):4139-4145. https://doi.org/10.1039/c9lc00721k. euploid embryo transfers. J Assist Reprod Genet. 2021;38(7):
90. Khosravi P, Kazemi E, Zhan Q, et al. Deep learning enables 1647-1653. https://doi.org/10.1007/s10815-021-02203-3.
robust assessment and selection of human blastocysts after 105. Geller J, Collazo I, Pai R, et al. An artificial intelligence-based
in vitro fertilization. NPJ Digit Med. 2019;2:21. https://doi.org/ algorithm for predicting pregnancy success using static images
10.1038/s41746-019-0096-y. captured by optical light microscopy during intracytoplasmic

n n
530 Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007
www.mcpdigitalhealth.org
CLINICAL DECISION SUPPORT SYSTEMS IN INFERTILITY

sperm injection. J Hum Reprod Sci. 2021;14(3):288-292. machine learning. Hum Reprod. 2023;38(10):1918-1926.
https://doi.org/10.4103/jhrs.jhrs_53_21. https://doi.org/10.1093/humrep/dead163.
106. Giscard d’Estaing S, Labrune E, Forcellini M, et al. A machine 122. Herbert S-L, Staib C, Wallner T, et al. Morphokinetic analysis
learning system with reinforcement capacity for predicting the of early human embryonic development and its relationship to
fate of an ART embryo. Syst Biol Reprod Med. 2021;67(1):64- endometriosis resection: a retrospective time-lapse study us-
78. https://doi.org/10.1080/19396368.2020.1822953. ing the KIDScoreTM D3 and D5 implantation data algorithm.
107. Lee C-I, Su Y-R, Chen C-H, et al. End-to-end deep learning for Arch Gynecol Obstet. 2023;308(2):587-597. https://doi.org/10.
recognition of ploidy status using time-lapse videos. J Assist 1007/s00404-023-07008-6.
Reprod Genet. 2021;38(7):1655-1663. https://doi.org/10. 123. Zieli
nski K, Pukszta S, Mickiewicz M, et al. Personalized predic-
1007/s10815-021-02228-8. tion of the secondary oocytes number after ovarian stimula-
108. Liao Q, Zhang Q, Feng X, et al. Development of deep learning tion: A machine learning model based on clinical and
algorithms for predicting blastocyst formation and quality by genetic data. PLoS Comput Biol. 2023;19(4):e1011020.
time-lapse monitoring. Commun Biol. 2021;4(1):415. https:// https://doi.org/10.1371/journal.pcbi.1011020.
doi.org/10.1038/s42003-021-01937-1. 124. Ferraretti AP, La Marca A, Fauser BC, et al. ESHRE
109. Sawada Y, Sato T, Nagaya M, et al. Evaluation of artificial intel- consensus on the definition of ‘poor response’ to ovarian
ligence using time-lapse images of IVF embryos to predict live stimulation for in vitro fertilization: the Bologna criteria.
birth. Reprod Biomed Online. 2021;43(5):843-852. https://doi. Hum Reprod. 2011;26(7):1616-1624. https://doi.org/10.
org/10.1016/j.rbmo.2021.05.002. 1093/humrep/der092.
110. Targosz A, Przystałka P, Wiaderkiewicz R, Mrugacz G. Seman- 125. Cherouveim P, Velmahos C, Bormann CL. Artificial intelligence
tic segmentation of human oocyte images using deep neural for sperm selection-a systematic review. Fertil Steril. 2023;
networks. Biomed Eng Online. 2021;20(1):40. https://doi.org/ 120(1):24-31. https://doi.org/10.1016/j.fertnstert.2023.05.157.
10.1186/s12938-021-00864-w. 126. Andrabi SW, Ara A, Saharan A, Jaffar M, Gugnani N,
111. Thirumalaraju P, Kanakasabapathy MK, Bormann CL, et al. Esteves SC. Sperm DNA fragmentation: causes, evaluation
Evaluation of deep convolutional neural networks in classifying and management in male infertility. JBRA Assist Reprod.
human embryo images based on their morphological quality. 2024;28(2):306-319. https://doi.org/10.5935/1518-0557.
Heliyon. 2021;7(2):e06298. https://doi.org/10.1016/j.heliyon. 20230076.
2021.e06298. 127. Bulletti FM, Berrettini M, De Luca R, et al. Calculators of pre-
112. Ueno S, Berntsen J, Ito M, et al. Pregnancy prediction performance dictive results: empowering assisted reproductive technologies
of an annotation-free embryo scoring system on the basis of deep programs. Clin Med. 2023;5:1057.
learning after single vitrified-warmed blastocyst transfer: a single- 128. Sinai I, Igras S, Lundgren R. A practical alternative to calculating
center large cohort retrospective study. Fertil Steril. 2021;116(4): unmet need for family planning. Open Access J Contracept.
1172-1180. https://doi.org/10.1016/j.fertnstert.2021.06.001. 2017;8:53-59. https://doi.org/10.2147/OAJC.S137705.
113. Xi Q, Yang Q, Wang M, et al. Individualized embryo selection 129. Infertility. CDC; April 26, 2023. https://www.cdc.gov/repro
strategy developed by stacking machine learning model for ductive-health/infertility-faq/?CDC_AAref_Val¼https://www.cd
better in vitro fertilization outcomes: an application study. c.gov/reproductivehealth/infertility/index.htm. Accessed April 4,
Reprod Biol Endocrinol. 2021;19(1):53. https://doi.org/10. 2024.
1186/s12958-021-00734-z. 130. Huang J, Lv P, Lian Y, et al. Construction of machine learning
114. Zhao M, Xu M, Li H, et al. Application of convolutional neural tools to predict threatened miscarriage in the first trimester
network on early human embryo segmentation during in vitro based on AEA, progesterone and b-hCG in China: a multi-
fertilization. J Cellular Molecular Med. 2021;25(5):2633-2644. centre, observational, case-control study. BMC Pregnancy
https://doi.org/10.1111/jcmm.16288. Childbirth. 2022;22(1):697. https://doi.org/10.1186/s12884-
115. Ahlström A, Lundin K, Lind A-K, et al. A double-blind random- 022-05025-y.
ized controlled trial investigating a time-lapse algorithm for 131. Guo X, Zhan H, Zhang X, et al. Predictive models for start-
selecting Day 5 blastocysts for transfer. Hum Reprod. 2022; ing dose of gonadotropin in controlled ovarian hyperstimu-
37(4):708-717. https://doi.org/10.1093/humrep/deac020. lation: review and progress update. Hum Fertil (Camb). 2023;
116. Berntsen J, Rimestad J, Lassen JT, Tran D, Kragh MF. Robust and 26(6):1609-1616. https://doi.org/10.1080/14647273.2023.
generalizable embryo selection based on artificial intelligence 2285937.
and time-lapse image sequences. PLoS One. 2022;17(2): 132. Kragh MF, Karstoft H. Embryo selection with artificial intelli-
e0262661. https://doi.org/10.1371/journal.pone.0262661. gence: how to evaluate and compare methods? J Assist Reprod
117. Diakiw SM, Hall JMM, VerMilyea M, et al. An artificial intelli- Genet. 2021;38(7):1675-1689. https://doi.org/10.1007/
gence model correlated with morphological and genetic fea- s10815-021-02254-6.
tures of blastocyst quality improves ranking of viable 133. Kirkegaard K, Agerholm IE, Ingerslev HJ. Time-lapse monitoring
embryos. Reprod Biomed Online. 2022;45(6):1105-1117. as a tool for clinical embryo assessment. Hum Reprod. 2012;
https://doi.org/10.1016/j.rbmo.2022.07.018. 27(5):1277-1285. https://doi.org/10.1093/humrep/des079.
118. Enatsu N, Miyatsuka I, An LM, et al. A novel system based on 134. Kirkegaard K, Ahlström A, Ingerslev HJ, Hardarson T.
artificial intelligence for predicting blastocyst viability and visu- Choosing the best embryo by time lapse versus standard
alizing the explanation. Reprod Med Biol. 2022;21(1):e12443. morphology. Fertil Steril. 2015;103(2):323-332. https://doi.
https://doi.org/10.1002/rmb2.12443. org/10.1016/j.fertnstert.2014.11.003.
119. Bamford T, Easter C, Montgomery S, et al. A comparison of 12 135. Leung ETY, Lee C-L, Tian X, et al. Simulating nature in sperm
machine learning models developed to predict ploidy, using a selection for assisted reproduction. Nat Rev Urol. 2022;19(1):
morphokinetic meta-dataset of 8147 embryos. Hum Reprod. 16-36. https://doi.org/10.1038/s41585-021-00530-9.
2023;38(4):569-581. https://doi.org/10.1093/humrep/dead034. 136. Afnan MAM, Liu Y, Conitzer V, et al. Interpretable, not black-
120. Barnes J, Brendel M, Gao VR, et al. A non-invasive artificial in- box, artificial intelligence should be used for embryo selection.
telligence approach for the prediction of human blastocyst Hum Reprod Open. 2021;2021(4):hoab040. https://doi.org/10.
ploidy: a retrospective model development and validation 1093/hropen/hoab040.
study. Lancet Digit Health. 2023;5(1):e28-e40. https://doi.org/ 137. Costa-Borges N, Munné S, Albó E, et al. First babies conceived
10.1016/S2589-7500(22)00213-8. with automated intracytoplasmic sperm injection. Reprod Bio-
121. Ferrand T, Boulant J, He C, et al. Predicting the number of oo- med Online. 2023;47(3):103237. https://doi.org/10.1016/j.
cytes retrieved from controlled ovarian hyperstimulation with rbmo.2023.05.009.

n n
Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007 531
www.mcpdigitalhealth.org
 MAYO CLINIC PROCEEDINGS: DIGITAL HEALTH

138. Lehman CD, Wellman RD, Buist DSM, et al. Diagnostic accuracy software-medical-device-samd/artificial-intelligence-and-machine-
of igital screening mammography with and without computer- learning-software-medical-device. Accessed April 5, 2024.
aided detection. JAMA Intern Med. 2015;175(11):1828-1837. 140. Habbema JDF, Eijkemans MJC, Leridon H, te Velde ER. Real-
https://doi.org/10.1001/jamainternmed.2015.5231. izing a desired family size: when should couples start? Hum
139. FDA. Artificial intelligence and machine learning in software as a Reprod. 2015;30(9):2215-2221. https://doi.org/10.1093/
medical device. FDA.. https://www.fda.gov/medical-devices/ humrep/dev148.

n n
532 Mayo Clin Proc Digital Health December 2024;2(4):518-532 https://doi.org/10.1016/j.mcpdig.2024.08.007
www.mcpdigitalhealth.org